Hey
I am Shivam
Software Developer / Data Scientist
Let me Introduce Myself
Hello Everyone. I am Shivam Sharma, Final year student pursuing Electronics and Communication Engineering from Maulana Azad National Institute of Technology Bhopal.
Though my discipline is Electronics, Machine Learning and Software Development have always been my keen interests. I would love to showcase my skills if given an opportunity.
Domain Specialization
From Software Development to Data Science, you can get solutions for a long range of problems.
Software Developer
Proficient with Data Structure and Algorithms and familiar with few development technologies.
Data Scientist
Working with ML models for predictive analysis and deep learning models for big problems
Data Analyst
Pandas for data analysis and Matplotlib/Seaborn to plot the data to gain meaningful insights
SKILLS
Projects
These are some of my best projects which were completed during my tenure at college.
Online Tshirt Store
This Online Tshirt Store is a dynamic website built using MERN stack (MongoDB, Express, React and Node) and implements the CRUD (Create, Read, Update, Delete) functions. Stripe payment gateway has been integrated inside the system for easy payments.It has two portals namely, Admin and User. The admin has the power to manipulate the database. The database used is MongoDB and it has a separate user profile for each user.
Eat or Beat
This project aims to provide suggestions on food items based on previous dataset obtained through a fitness device like fitbit, using an android application.In App we have two options to send query to server, first is scanning the barcode of the product and then using speech recognition, for which we use Google Speech Recognition API. The output displays whether the food item is suitable or not along with a suggestion for better results.
Obstacle Detection and Navigation
This projects works on the ssd_inception_v2 (fast and accurate) provided by Tensorflow object detection API to detect the obstacles. The algorithm is implemented with the help of OpenCV to command the user to move in the direction in order to avoid the obstacles. It is a standalone application and can be used anywhere by connecting with a server (firebase, AWS etc) and can be used with any android or web application.
Time Series Prediction
This project aims to find the future predictions of a time series which could not be predicted using regression techniques. This model utilizes a RNN which uses a LSTM (Long short term memory) which finds seasonality and periodicity of the time series with minimum MAE (mean absolute error).
Multi Object Classifier
The Multi Object classifier uses the MNIST dataset to correctly classify different objects. It also increases the number of images in the dataset by using image augmentation techniques. The Model is trained using a convolutional neural network and it achieves accuracy of about 85% on the validation data set.
Neural Machine Translation
The Neural Machine Translation will input a date written in a variety of possible formats and translate them into standardized, machine readable dates. It utilizes the attention mechanism to tell the model where it should pay attention to at any step. It utilizes two separate bidirectional LSTMs to train the model.